fix(VAPI-4457): scope SSML un-escape to SpeakSentence text

bandwidth/models/bxml/nestable_verb.py

@@ -6,7 +6,6 @@
 @copyright Bandwidth INC
 """
 from __future__ import annotations
-import re
 from typing import Union
 import xml.etree.ElementTree as ET
 
@@ -16,7 +15,6 @@
 class NestableVerb(Verb):
     """Base class for nestable BXML verbs
     """
-    ssml_regex = r"<([a-zA-Z//].*?)>"
 
     def __init__(self, tag: str, content: str = None, nested_verbs: list[Verb] = None):
         """Initialize the verb model

bandwidth/models/bxml/root.py

@@ -5,17 +5,15 @@
 
 @copyright Bandwidth INC
 """
-import re
 from typing import List
 import xml.etree.ElementTree as ET
 
-from bandwidth.models.bxml.verb import Verb
+from bandwidth.models.bxml.verb import Verb, _unescape_ssml
 
 
 class Root:
     """Base class for BXML roots
     """
-    ssml_regex = r"<([a-zA-Z//].*?)>"
 
     def __init__(self, tag: str, nested_verbs: List[Verb] = None):
         """Initialize instance of class
@@ -76,4 +74,4 @@ def to_bxml(self) -> str:
             str: Serialized BXML string
         """
         xml_document = self._generate_xml()
-        return re.sub(self.ssml_regex, r"<\1>", ET.tostring(xml_document._root, encoding='UTF-8', method='xml', xml_declaration=True).decode("utf8"))
+        return _unescape_ssml(ET.tostring(xml_document._root, encoding='UTF-8', method='xml', xml_declaration=True).decode("utf8"))

bandwidth/models/bxml/verb.py

@@ -10,6 +10,19 @@
 from typing import Union
 import xml.etree.ElementTree as ET
 
+_SPEAK_SENTENCE_REGEX = re.compile(r"(<SpeakSentence\b[^>]*(?<!/)>)(.*?)(</SpeakSentence>)", re.DOTALL)
+_SSML_REGEX = re.compile(r"&lt;([a-zA-Z//].*?)&gt;")
+
+
+def _unescape_ssml(bxml: str) -> str:
+    """Restore SSML tags inside SpeakSentence text content only.
+
+    Attribute values and all other elements keep their XML escaping.
+    """
+    return _SPEAK_SENTENCE_REGEX.sub(
+        lambda m: m.group(1) + _SSML_REGEX.sub(r"<\1>", m.group(2)) + m.group(3), bxml
+    )
+
 
 class Verb:
     """Base class for BXML verbs
@@ -93,8 +106,5 @@ def to_bxml(self) -> str:
         Returns:
             str: Serialized BXML string
         """
-        ssml_regex = r"&lt;([a-zA-Z//].*?)&gt;"
-
         xml_document = self._generate_xml()
-        # if the root name is speakSentence, replace content with the SSML regex
-        return re.sub(ssml_regex, r"<\1>", ET.tostring(xml_document._root).decode('utf8'))
+        return _unescape_ssml(ET.tostring(xml_document._root).decode('utf8'))

bandwidth/models/bxml/verbs/speak_sentence.py

@@ -5,10 +5,6 @@
 
 @copyright Bandwidth INC
 """
-import re
-
-import xml.etree.ElementTree as ET
-
 from ..nestable_verb import NestableVerb
 
 
@@ -44,7 +40,3 @@ def _attributes(self):
             "gender": self.gender,
             "locale": self.locale,
         }
-
-    def to_bxml(self) -> str:
-        xml_document = self._generate_xml()
-        return re.sub(self.ssml_regex, r"<\1>", ET.tostring(xml_document._root).decode('utf8'))

test/unit/models/bxml/test_gather.py

@@ -69,3 +69,11 @@ def test_add_verb(self):
         expected = '<Gather gatherUrl="test.com" gatherMethod="POST" gatherFallbackUrl="fallback-test.com" gatherFallbackMethod="GET" username="user" password="pass" fallbackUsername="user" fallbackPassword="pass" tag="tag" terminatingDigits="2" maxDigits="5" interDigitTimeout="1" firstDigitTimeout="3" repeatCount="2" input="dtmf_speech" hints="yes,no" language="en-US" partialResultCallback="partial-test.com" partialResultCallbackMethod="POST" profanityFilter="true" speechModel="default" speechTimeout="4"><PlayAudio username="user" password="pass">https://audio.url/audio1.wav</PlayAudio><SpeakSentence>Hello. Your number is <say-as interpret-as="telephone">asdf</say-as>, lets play a game. What is 10 + 3. Press the pound key when finished.</SpeakSentence></Gather>'
         self.gather.add_verb(self.speak_sentence)
         assert expected == self.gather.to_bxml()
+
+    def test_ssml_unescaped_in_speak_sentence_but_not_in_gather_attributes(self):
+        expected = '<Gather gatherUrl="https://example.com/?a=&lt;b&gt;"><SpeakSentence>say <emphasis>this</emphasis></SpeakSentence></Gather>'
+        gather = Gather(
+            gather_url="https://example.com/?a=<b>",
+            audio_verbs=[SpeakSentence(text="say <emphasis>this</emphasis>")]
+        )
+        assert expected == gather.to_bxml()

test/unit/models/bxml/test_response.py

@@ -7,7 +7,7 @@
 """
 import unittest
 
-from bandwidth.models.bxml import Response
+from bandwidth.models.bxml import Response, Redirect, SpeakSentence
 
 
 class TestResponse(unittest.TestCase):
@@ -23,3 +23,25 @@ def test_response_init(self):
         """
         expected_bxml = "<?xml version='1.0' encoding='UTF-8'?>\n<Response />"
         assert self.response.to_bxml() == expected_bxml
+
+    def test_ssml_unescaped_but_other_attributes_untouched(self):
+        """SSML in <SpeakSentence> text is un-escaped; escaped characters in other elements' attributes are not
+        """
+        expected_bxml = "<?xml version='1.0' encoding='UTF-8'?>\n<Response><Redirect redirectUrl=\"https://example.com/a?info=&lt;https://example.com/c.pem&gt;\" /><SpeakSentence voice=\"julie\" gender=\"female\" locale=\"en_US\">Hello. Your number is <say-as interpret-as=\"telephone\">asdf</say-as>, lets play a game.</SpeakSentence></Response>"
+        self.response.add_verb(Redirect(redirect_url="https://example.com/a?info=<https://example.com/c.pem>"))
+        self.response.add_verb(SpeakSentence(
+            text='Hello. Your number is <say-as interpret-as="telephone">asdf</say-as>, lets play a game.',
+            voice="julie",
+            gender="female",
+            locale="en_US"
+        ))
+        assert expected_bxml == self.response.to_bxml()
+
+    def test_empty_speak_sentence_does_not_swallow_following_verbs(self):
+        """A <SpeakSentence> with no text must not capture the verbs that follow it
+        """
+        expected_bxml = "<?xml version='1.0' encoding='UTF-8'?>\n<Response><SpeakSentence /><Redirect redirectUrl=\"https://example.com/a?info=&lt;b&gt;\" /><SpeakSentence>hi <break/></SpeakSentence></Response>"
+        self.response.add_verb(SpeakSentence(text=""))
+        self.response.add_verb(Redirect(redirect_url="https://example.com/a?info=<b>"))
+        self.response.add_verb(SpeakSentence(text="hi <break/>"))
+        assert expected_bxml == self.response.to_bxml()